Harris County, Texas--comprised of portions of the Houston metropolitan area--was once justifiably called the "epicente" of the death penalty in the United States. (1) This status was cemented during the peak of the post-Furman era (2) of death sentencing in the early-to-mid 1990's. (3) In 1994 seventeen defendants were sentenced to death in Harris County. (4) Yet twenty years later in 2014 only four defendants were sentenced to death there. (5) This plunge is startling. But it is also almost exactly representative of the United States as a whole, where death sentences plunged from 310 in 1994 to 73 in 2014--an astonishing 76% decline. (6)
The popular media have posited a variety of reasons for this decline. (7) So have scholars. But existing scholarly analyses--while illuminating--are either without extensive empirical analysis, (8) or empirical analysis that is limited to certain jurisdictions or time periods, (9) or nationwide in scope, but without examination of the details of individual cases. (10) This article reports the results of an empirical analysis that spans the whole country over a two-decade period at the level of individual case characteristics, in search of significant contributing factors in the decline of death sentences.
The article proceeds in three parts. Part I explains the methodology for unearthing relevant data and preparing it for analysis. Part II analyzes the decline in death sentences due to decreasing death eligibility, that is, fewer murderers over time who meet the criteria that made death a sentencing option. Four reasons will be examined: fewer death-eligible murders, (11) the United States Supreme Court's exemptions of juveniles who were less than eighteen years of age at the time of the commission of the murder (12) and persons with intellectual disability (known to the law as the "mentally retarded"), (13) and the abolition of the death penalty in several states. (14) Finally, Part III examines increasingly narrower perceptions of death-worthiness, that is, the evolution in attitudes among prosecutors and sentencers (15) toward deeming fewer among the many death-eligible defendants worthy of death sentences. This Part requires the most complicated analysis because unlike death-eligibility decisions, which are dictated by law, (16) death-worthiness decisions emerge from an opaque brew of many factors, including, but not limited to, resource differentials among jurisdictions, (17) prosecutorial attitudes, (18) the wishes of the murder victim's survivors, (19) defense counsel performance, (20) public opinion, (21) and sentencer reactions. (22) But while death-worthiness decisions are often opaque in individual cases, each case generates empirical data from which patterns may be discerned. Part III uses such data to analyze ten questions:
* Did the advent of life-without-parole ("LWOP") reduce death sentences in jurisdictions where it was added as an option?
* Did sentencers become more reluctant to return death sentences?
* Were death sentences decreasingly imposed in less aggravated cases and increasingly imposed in more aggravated cases?
* Did presentation of greater numbers of mitigating factors conduce to fewer death sentences?
* Did robbery during a murder became a less powerful aggravator?
* Did 18- to 20-year-olds benefit from a ripple effect from the exemption of juveniles?
* Did death sentences become less common in multiple perpetrator cases?
* Did low population counties increasingly drop out of death sentencing?
* Did low revenue counties increasingly drop out of death sentencing? and
* Did a few traditionally high-volume death sentencing counties skew the figures by cutting back on the use of the death penalty due to local political factors?
We did not have the resources to gather and analyze data from every year in the post-Furmam era. Instead, we decided to focus on three particular years: 1994, 2004, and 2014--a monumental enough undertaking in itself. We chose 1994 as the starting point because it was one of the peak death sentencing years in the post-Furman era, (23) and because 1994 enabled us to leap two decades ahead to the most recent year for which data were available when we began the project--2014. Then 2004 became the midpoint for sampling what happened between 1994 and 2014. The data collection for those three years proceeded in four steps.
Thorough analysis required inclusion of all of the death-sentenced cases, and a robust control group death-eligible cases in which defendants did not receive death sentences. Thus, step one was to find as many death-eligible cases as possible. A case was considered death-eligible if any of nineteen statutory aggravating circumstances that typically render a murder death-eligible were present: multiple murders; murders during robbery, rape, kidnapping, home burglary, or arson; to escape incarceration; for financial motive other than robbery; to eliminate or retaliate against a witness; for an anti-government or terroristic motive; hate crime; prior murder conviction; prior violent felonies; committing the murder while incarcerated; victim twelve years of age or younger; victim seventy years of age or older; victim a government servant (usually a police officer); torture; and hiring a killer/acting as a hired killer. (24)
We created three designations: Death-Sentenced ("DS"), Sentencer-Spared ("SS") (the sentencer was presented with the option of death but chose a lesser sentence), (25) and Prosecutor-Spared ("PS") (the prosecutor did not pursue a death sentence through the sentencing phase, and thus a lesser sentence was imposed). (26) DS cases were relatively easier to find by comparing the lists from succeeding quarterly Death Row USA reports (27) name-by-name--over 3,000 names for each quarter. (28) For SS and PS cases no equivalent lists existed, so these cases were identified through news database searches. (29) The only way not to exclude relevant cases was to use the very broad search term "death/s sentence or penalty." This returned tens of thousands of articles for each year, only a small proportion of which were relevant.
The final tally of 1665 death-eligible cases, comprised of 517 DS cases, 311 SS cases, and 837 PS cases, is shown in Table 1, below. In summary then, the database includes every DS case (30) from the three years along with very robust comparison sets of SS and PS cases-indeed, the SS and PS cases outnumber the DS cases 1148 to 517. (31)
This Table and Tables 2 through 14 throughout the remainder of the article are all available in graph form and in color in the Online Table-to-Graph Appendix. (33)
Step two was to find as many details as possible about each case. The two primary sources were online news databases for all three years (34) and appellate opinions for 1994 and 2004 DS cases. (35) Appellate opinions did not yet exist for 2014 DS cases, so in addition to news reports, we ascertained as many names as we could of the prosecutors and defense lawyers in each case, and sent each of them a letter inquiring about the facts of their cases. We received a fair number of replies. (36)
Step three was to code each case for scores of variables, including factors relating to where the case was litigated (state or federal court, and if in state court, in which county, the county's population as of the most applicable census, (37) and the county's revenue for the applicable year (38)), age of the defendant, sixty-six aggravating factors, (39) and twenty-two mitigating factors. (40) Then we input the coded variables into a Statistical Package for the Social Sciences ("SPSS") spreadsheet. (41)
Finally, step four was to analyze that data. That analysis is set forth in the remainder of this article. Much of the analysis is presented in the form of counting and percentages, which are techniques of descriptive statistics. (42)
DECREASES TO DEATH-ELIGIBILITY
Decrease in Death-Eligible Murders
The starting point for examining the decrease in death-eligible murders is the FBI's Uniform Crime Reports. (43) These Reports list the number of "Murders and non-negligent manslaughters" ("murders") in each state for each year. (44) In the thirty-seven states that had the death penalty in 1994 (45) there were 19,250 murders in 1994, (46) which declined dramatically to 13,800 in 2004, (47) and further but less dramatically to 12,440 in 2014. (48) These figures include the six jurisdictions that abolished the death penalty after 2003, (49) but those effects will be subtracted shortly. (50)
There is no listing of non-death sentences in death-eligible cases for any year. The figures for all murders, however, would be useful to calculate the likely number of death-eligible murders if the relative proportion of death-eligible murders could be ascertained for the three years at issue. Fortunately, a remarkable empirical study gives good reason to believe that the proportion of murders that are death-eligible remains virtually constant over time. Fagan, Zimring, and Geller conducted a massive and painstakingly detailed study of over 490,000 murders in the United States from 1976 to 2003 in both death penalty and non-death penalty states. (51) They concluded:
The pattern of capital-eligible homicides [in the nondeath penalty states] fluctuates over time in a manner similar to the death penalty states.... [T]here is little variation in the rates of capital-eligible homicides over time [and] the shape of the temporal trends in capital-eligible homicides in death penalty states and nondeath penalty states is nearly identical. (52) Since death-eligible cases remained a constant proportion of murders, they decreased proportionately along with the overall decrease in murders. (53) Thus, it is possible to calculate how many death sentences there would likely have been in 2004 and 2014 if the same death-sentencing rate had prevailed as in 1994, (54) and nothing else had changed. (55) Note that...